Specialized AI course: Generative AI vs Specialized AI

[Stefan.He_Enlift (Stefan He)]

Hello,

I was going through the course “Introduction to Specialized AI” and was curious why the two AI types supported by UiPath are Generative AI and Specialized AI? And not something like General AI vs Specialized AI.

The reason I ask is because in my general understanding of the term “Generative AI”. It is used to generate something new. Images, Sentences or Audio from prompts.

From my current understanding of UiPaths product offerings, UiPath doesn’t offer Generative AI’s as a product, but rather integrations with existing Generative AI’s.

And while services like Communications mining do use specialized AI, services like Document Understanding uses a mix of general ai models (E.g. DocumentUnderstanding) and specialized ai models (e.g. Passports or Invoices) with the option to apply Generative AI’s through the services of external Generative AI providers (e.g. Chat-GTP). But this implies the use of General, Specialized and Generative.

The final part I’m a bit confused about is why LLM models are classified as Generative AI’s. While I believe that is “technically” true, It feels like a mis-representation of what Generative AI’s and LLM can do, in isolation and when combined together. From my understanding of the current UiPath offerings, I’m not sure when or how a LLM model is used to actually Generate new data.

So to summarize… why are AI’s in UiPath split into Specialized and Generative? And not more specific types.

[Krishna_Raj]

@Stefan.He_Enlift

Introduction to Specialized AI

check the thread

Generative AI - How its used or can be used

In the context of UiPath, there are potential areas where generative AI techniques can be utilized: Natural Language Processing (NLP): Generative AI can be used to generate human-like text, enabling more advanced chatbots, automated email responses, or natural language interaction in RPA processes. Intelligent Automation: Generative AI can assist in automating tasks that require creativity or decision-making based on past patterns. For example, generating personalized emails, creating dynamic …

[Stefan.He_Enlift (Stefan He)]

I understand that UiPath can be used with Generative AI and what Generative AI can be used to achieve. But UiPath as an Automation platform can integrate with any service that has a User interface or API that’s been exposed.

Wouldn’t it be more prudent to label the AI offerings as UiPath Out-Of-The-Box vs Upload your own model, if you want to talk about “Potential offerings”? Especially when the Generative AI use cases that UiPath offers are locked behind specialized AI’s or integrations with third party services?